Your data. Your chatbot. Your API.
Upload your documents, configure a knowledge base, and serve a chatbot that actually knows your business. No fine-tuning required — pure RAG, cited answers.
Trained on your data, no fine-tuning
Retrieval-augmented generation means your chatbot pulls answers from your live documents. Update a PDF, the chatbot learns instantly.
Cited, verifiable answers
Every response shows which file and page it came from. Critical for support, legal, sales, and compliance use cases.
Multi-model under the hood
Use Qwen3 for fast chat, DeepSeek for reasoning, Kimi for long context. One config, swap models per request.
OpenAI-compatible API
Embed it in any product — Slack bot, support widget, internal tool — using the OpenAI SDK you already know.
How it works
Create a knowledge base
One click. Free. Name it after the chatbot you want to build.
Upload documents
PDFs, text files. Up to 50 MB per file. Free 15 MB total on PAYG, 50 MB on Pro.
Configure the chatbot
Pick a model, set a system prompt. Done.
Call the API
POST to /v1/chat/completions with kb_id. Get cited answers back.
Build a chatbot that actually knows your business.
Stop relying on generic LLMs that hallucinate. Start with a $1 free credit, upload a doc, and have a working custom chatbot in minutes.
Related